undefined | Better HN

0 pointsdeathanatos5y ago0 comments

My understanding is that they took in RGB picture data, yes, but that they were required to be square (s.t. matrix transforms work on them, I think?).

But that's still different from taking in the actual JPEG data, which is sort of what the parent gets at: something has to decode that, and that software isn't a neural net.

(Further, when I worked w/ ML that dealt w/ image data, we had a host of non-ML code written around it to support it, dealing with the various facets of running in the cloud, where to get the data, where to store the results, who to notify about the results, and a bunch of preprocessing on the image — such as removing pointless borders that humans put around images.)

0 comments

2 comments · 2 top-level

recursivecaveat5y ago

Just FYI, non-square images should work fine. Arguably that's one of the selling points of convolutional neural nets. The early big-name training sets tended to be all square though, and if your design ends with a fully-connected layer (more common back then as well) then you're pretty locked-in for aspect ratio unless you retrain.

ajconway5y ago

That software can be generated and continuously optimized by a neutral net. But that would probably mean that we achieved AGI, so decoding JPEGs won’t be that exciting by itself.

j / k navigate · click thread line to collapse